Abstract
This paper introduces the methodology of domain knowledge exploration in so called rule-based knowledge bases from the medical perspective, but it could easily by transformed into any other domain. The article presents the description of the CluVis software with rules clustering and visualization implementation. The rules are clustered by using hierarchical clustering algorithm and the resulting groups are visualized using the tree maps method. The aim of the paper is to present how to explore the knowledge hidden in rule-based knowledge bases. Experiments include the analysis of the influence of different clustering parameters on the representation of knowledge bases.
This is a preview of subscription content, log in via an institution to check access.
Notes
- 1.
Age of the patient: (1) young, (2) pre-presbyopic, (3) presbyopic, spectacle prescription: (1) myope, (2) hypermetrope, astigmatic: (1) no, (2) yes and tear production rate: (1) reduced, (2) normal.
- 2.
1: hard contact lenses, 2: soft contact lenses and 3:no contact lenses. Class distribution is following: 1: 4, 2: 5 and 3: 15.
- 3.
There are many possible ways to define the stop condition. For example, it can reach the specified number of groups, or reach the moment when the highest similarity is under a minimal required threshold (which means that the groups of rules are now more differential than similar to one another).
- 4.
In this task clustering is stopped when given number of clusters is generated.
- 5.
If both compared objects have the same attribute and this attribute has the same value for both objects then add 1 to a given similarity measure. If otherwise, do nothing.
- 6.
IOF measure assigns a lower similarity to mismatches on more frequent values while the OF measure gives opposite weighting for mismatches when compared to the IOF measure, i.e., mismatches on less frequent values are assigned a lower similarity and mismatches on more frequent values are assigned a higher similarity.
- 7.
The most complex of the all used inter-cluster similarity measures as it handles numerical attributes and symbolic attributes differently.
- 8.
However, the authors see the necessity to analyze more methods for creating clusters’ representatives and their influence on the resultant structure efficiency.
- 9.
The i-th variable value is accessed by its name in map, not by its index.
- 10.
In this poarticular case, the authors have used the contact lenses dataset, Gower’s similarity measure and SL clustering method. The representative presented here is the description of the clusters J5 which contains 5 elements and the size of its representative is equal to 4.
- 11.
The meaning of the columns in Table 3 is as follows: U - number of singular clusters in the resultant structure of grouping, BRS - a biggest representative’s size - number of descriptors used to describe the longest representative, ARS - an average representative’s size - an average number of descriptors used to describe cluster’s representatives, wARS - a weighted average representative’s size (Attributes) - a quotient of an average number of descriptors used to describe cluster’s representative in a given data set and the number of attributes in this dataset, BRL - a biggest representative’s length - the number of descriptors in a biggest cluster’s representative and BCS - a biggest cluster’s size - number of rules in the cluster that contains the most of them. Clusters is the number of the created clusters of rules while Nodes is the number of nodes in the dendrogram representing the resultant structure.
References
Bazan, J.G., Szczuka, M.S., Wróblewski, J.: A new version of rough set exploration system. In: Alpigini, J.J., Peters, J.F., Skowron, A., Zhong, N. (eds.) RSCTC 2002. LNCS, vol. 2475, pp. 397–404. Springer, Heidelberg (2002). doi:10.1007/3-540-45813-1_52
Boriah, S., Chandola, V., Kumar, V.: Similarity measures for categorical data: a comparative evaluation. In: Proceedings of the 8th SIAM International Conference on Data Mining, pp. 243–254 (2008)
Dubes, R., Jain, A.K.: Clustering techniques: the user’s dilemma. Pattern Recognit. 8(4), 247–260 (1976)
Lichman M. UCI Machine Learning Repository, University of California (2013). http://archive.ics.uci.edu/ml
Nowak-Brzezińska, A.: Mining rule-based knowledge bases inspired by rough set theory. Fundamenta Informaticae 148, 35–50 (2016). doi:10.3233/FI-2016-1421. IOS Press
Przybyła-Kasperek, M., Wakulicz-Deja, A.: The strength of coalition in a dispersed decision support system with negotiations. Eur. J. Oper. Res. 252, 947–968 (2016)
Simiński, R.: Multivariate approach to modularization of the rule knowledge bases. In: Gruca, A., Brachman, A., Kozielski, S., Czachórski, T. (eds.) Man–Machine Interactions 4. AISC, vol. 391, pp. 473–483. Springer, Cham (2016). doi:10.1007/978-3-319-23437-3_40
Grzymala-Busse, J.W.: A new version of the rule induction system LERS. Fundamenta Informaticae 31, 27–39 (1997)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2017 Springer International Publishing AG
About this paper
Cite this paper
Nowak-Brzezińska, A., Rybotycki, T., Simiński, R., Przybyła-Kasperek, M. (2017). Knowledge Exploration in Medical Rule-Based Knowledge Bases. In: Nguyen, N., Papadopoulos, G., Jędrzejowicz, P., Trawiński, B., Vossen, G. (eds) Computational Collective Intelligence. ICCCI 2017. Lecture Notes in Computer Science(), vol 10449. Springer, Cham. https://doi.org/10.1007/978-3-319-67077-5_15
Download citation
DOI: https://doi.org/10.1007/978-3-319-67077-5_15
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-67076-8
Online ISBN: 978-3-319-67077-5
eBook Packages: Computer ScienceComputer Science (R0)